Neutron activation analysis of multiple 10–100 μg glass samples from siderophile element partitioning experiments

Abstract

Experimental studies of the metal/silicate partitioning of Ir have produced two sets of results differing by a factor of 10 6 ( Jones and Drake, 1986; O'Neill et al., 1995). In an attempt to understand this discrepancy, we have performed instrumental neutron activation analyses of multiple glass chips from these siderophile element partitioning experiments. From seven to ten individual chips (~ 10–100 μg each) were analyzed from each experimental glass. The major target was Ir, which was detected at concentrations ranging from 690 ppb (ng/g) to less than 1 ppb. Detection limits are less than 10 −13 grams Ir. Almost all of the run products appear to be homogeneous with respect to lithophile elements, but most vary considerably in their Ir concentrations. It appears that most of the experiments analyzed so far contain small amounts of metal in the glass separates. One experiment from Jones and Drake (1983) appears to be homogeneous, but in fact may not be. More recent experiments are definitely not homogeneous, and require the existence of a metallic phase distinctly different from the bulk solid metal or liquid metal/sulfide phases also present in the charges. Iridium contents of the glass in the stirred crucible experiment studied ( O'Neill et al., 1996) decreased continuously over a period of several months, even as the oxygen fugacity was increasing, suggesting that the experiment never achieved equilibrium. Iridium contents in multiple samples of the glasses varied over factors of 2 to 3x. Further subdivision of one low-Ir glass did not result in a reduction in the variation of Ir between aliquots. Thus, the variation in Ir concentration appears to remain fairly constant, irrespective of the scale of the sampling.